Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    the words "hip hop" or job titles containing "vice president.". They are distinct enough that perhaps the model is using "hip" alone to find "hip hop" and "vice" alone to find job titles

    oai_token-act-pair · gemini-2.0-flash

    hip-hop

    np_max-act · gemini-2.0-flash
    New Auto-Interp
    Negative Logits
    ReusableCell
    -0.93
     resourceCulture
    -0.89
     CreateTagHelper
    -0.79
    RetentionPolicy
    -0.74
    تقاوى
    -0.74
     kaynağından
    -0.71
    GrantedAuthority
    -0.70
     HasFactory
    -0.69
     JpaRepository
    -0.68
    TableHead
    -0.68
    POSITIVE LOGITS
    <bos>
    1.84
    '
    0.50
    @
    0.50
    x
    0.47
    u
    0.47
    t
    0.46
    H
    0.45
    Enter
    0.45
    di
    0.45
    is
    0.45
    Act Density 0.187%

    No Known Activations